The series of articles contains a comparison of the use of information on building zones from three sources for dasymetric population mapping: from the Corine Land Cover project (CLC), from the Urban Atlas project (UA) and from the object classifi cation (OBIA) of the RapidEye data. These sources are characterized by varying spatial and thematic accuracy as well as a diff erent methodology of building separation. The experiment was carried out in the area of Kraków, using statistical data from 141 urban units (u.u.) of the city.In the fi rst part of the cycle, population conversions were presented based on the Corine Land Cover (CLC) and Urban Atlas (UA) databases. The second part presents the methodology of mapping construction zones, divided into several categories, by means of object classifi cation (OBIA). The classifi cations were carried out on four RapidEye satellite images. The developed map is the basis for the population calculation in three variants: binary method, and two surface-weight aggregation methods, where the proportions of population density for diff erent building categories are calculated by minimizing square error (RMSE) and percentage (MAPE) in census units. The obtained results of the population distribution indicate the need to determine the function of development. Therefore, in addition, experiments were carried out combining OBIA results with the LULC map of the UA project. From the experiments it appears that from the tested six variants of population mapping the best is the surface-weight method based on OBIA+UA (RMSE = 4,270 people/u.u., MAPE = 75%.). Binary method based on OBIA+UA results at RMSE = 4540 people/u.u., MAPE = 108%. Results with the use of OBIA, without correction of building functions with UA, are incorrect (RMSE: 5958-7987 people/u.u., MAPE: 2262%-6612 %).In the subsequent parts of the publication cycle, the results obtained so far will be compared: three CLC-based maps, three UA-based maps, six maps based on OBIA / OBIA+UA. To verify the population map, a detailed reference map of the Bronowice district will be used as well as a 1-kilometer GUS grid. A discussion will be conducted related to the use of RMSE and MAPE parameters in the process of results optimization. A ranking of methods and recommendations will be developed to improve the results of population conversion based on CLC, UA and OBIA.Słowa kluczowe: dane demografi czne, modelowanie dazymetryczne, klasyfi kacja obiektowa, RapidEye
StreszczenieCykl artykułów zawiera porównanie możliwości wykorzystania do kartowania ludności informacji o strefach zabudowy z trzech źródeł: z projektu Corine Land Cover (CLC), z projektu Urban Atlas (UA) oraz z wyniku klasyfi kacji obiektowej (OBIA) danych RapidEye. Źródła te charakteryzują się różną dokładnością przestrzenną i tematyczną oraz różną metodologią wyodrębniania zabudowy. Eksperyment przeprowadzono na obszarze Krakowa, wykorzystując dane statystyczne ze 141 jednostek urbanistycznych miasta.W pierwszej części cyklu zaprezentowano przeliczanie populacji w oparciu...
